...is the PIH.
To a large extent that's an empirical question, though, and it doesn't unequivocally come out against naïve Keynesianism in the data.
And even on the theoretical side when you build in some kind of expectations in a New Keynesian framework you get all the Keynesian cross logic back even with PIH. You are essentially modernizing the Keynesian cross in the exact same way that Krugman modernized the old naïve liquidity trap. This, I think, is why many of us New Keynesian types are very comfortable saying that Old Keynesianism isn't all that bad - because a lot of the Old Keynesian results pop up in New Keynesian models, so long as keep expectations always in mind. So for the purpose of communicating the basic idea Old Keynesianism still works pretty well.
And if you read Keynes on expectations you realize that he probably had something like the New Keynesian approach in mind anyway, he just didn't get the formal work past the old naïve Keynesian model (and even that he left to Hicks, Samuelson, etc.).
Alright, if you'll excuse me I am going to duck the tomatoes that the Post Keynesians are throwing at me now.